Locus control region elements HS1 and HS4 enhance the therapeutic efficacy of globin gene transfer in β-thalassemic mice

The β-thalassemias are inherited autosomal recessive anemias caused by mutations that diminish or abolish expression of the β-globin gene.^1,2  The only current means to cure the disease is allogeneic bone marrow transplantation (BMT).^3,4  In the absence of a histocompatible donor, however, the genetic correction of autologous hematopoietic stem cells (HSCs) represents a highly attractive alternative treatment.⁴ 

Achieving therapeutic expression of the human β-globin transgene in hematopoietic chimeras has long posed major challenges in terms of transgene regulation, vector stability and transduction efficiency (reviewed in Persons and Tisdale,⁵  May and Sadelain,⁶  and Bank et al⁷ ). A careful selection of proximal and distal β-globin transcription control elements and the use of a recombinant HIV-1 genome to generate a stable vector eventually enabled May et al to cure β-thalassemia in mice.⁸  Their vector, termed TNS9,⁸  and all subsequent therapeutic globin vectors published to date, express the globin transgene under the control of the β-globin promoter and the HS2, 3, and 4 locus control region (LCR) elements.^8,,,,,,–15  The HS2 and HS3 elements are the most powerful single elements within the LCR (reviewed in Li et al¹⁶ ). The relative importance of HS1 and HS4 is less well defined.¹⁶ 

In order to select a vector for clinical evaluation in patients with β-thalassemia, we have therefore undertaken a quantitative analysis of the contribution of the β-globin HS1 and HS4 elements to the specificity, inducibility, long-term in vivo expression, and therapeutic potential of globin lentiviral vectors. It is indeed central to the success of safe stem cell engineering that therapeutic transgene expression be achieved with a low (ideally 1 to 2) vector copy number per cell.^17,18  We show here that HS1 and HS4 are major contributors to achieving therapeutic globin expression in β-thalassemic mice.

All vectors used in this study were derived from the previously described TNS9.⁸  Detailed description of vector construction and production is provided in Document S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article). Vector copy number quantification was performed by Southern blot as previously described⁸  and by TaqMan analysis (primers and probes appear in Document S1). MEL cell transduction and differentiation were done as previously described.⁸  For β-globin transgene analyses, total RNA was extracted from MEL cells or total mouse peripheral blood (PB) using TRIzol reagent (Invitrogen, Carlsbad, CA). Quantitative primer extension assays were performed as described previously.⁸  Hbb^th3/+ bone marrow (BM) chimeras were generated as described.^8,19  BM cells were prestimulated for 12 hours, transduced at 10⁶ cells/mL per well at a multiplicity of infection (MOI) of 20 to 35 for 8 hours and injected into lethally irradiated recipients (5 × 10⁵ to 10⁶ cells per recipient). A detailed description of the transduction procedure appears in Document S1. At several time points after bone marrow transplantation, PB was collected and hemoglobin levels were measured on a Coulter A^cT diff. instrument (Beckman Coulter, Brea, CA). Red cell lysates of freshly collected PB were analyzed by cellulose acetate electrophoresis (pH 8.5, Helena Laboratories, Beaumont, TX and quantified as described⁸  using ImageJ 1.38x software (http://rsb.info.nih.gov/ij/). Statistical analysis was done by the Student t test using SigmaStat 2.03.0 software (Access Softek, San Rafael, CA).

To evaluate the contribution of the 5′HS1 LCR element to lentivirus-encoded human β-globin expression, we created vectors encoding either HS2–3-4 or HS1–2-3–4 (Figure 1Ai). In MEL cells, addition of 5′HS1 had no effect on average transgene expression per vector copy in one vector (T9 vs T10, P = .669, Figure 1Aii), and even significantly decreased the average transgene expression per vector copy in another (S9 vs S10, P < .001). In stark contrast, addition of the 5′HS1 element significantly improved the vectors' performance in vivo (S9 vs S10 and T9 vs T10, Figure 1C). Globin transgene expression, at the mRNA level and per vector copy, increased from 27% (± 6%, mean ± SD, S9 and T9) of endogenous β-globin mRNA to 41% (± 9%, S10 and T10, P < .001). The average copy number ranged from 0.42 for T10 to 0.71 for S9 and remained stable throughout the experiment (Figure 1B). The mRNA data were confirmed on the protein level. All animals (n = 73) showed sustained amelioration of their anemia, without decrease over the period of observation. The T9 and S9 vectors conferred similar levels of chimeric hemoglobin (cHb, mα₂:hβ₂), on the order of 44% to 54% per vector copy and normalized to endogenous murine Hb, while vectors containing 5′HS1 element (S10 and T10) generated even higher levels of cHb (Figure 1D).

We next examined the contribution of the flanking regions of the HS4 element, which is a staple of TNS9⁸  and other therapeutic globin vectors.^{9,–11,13,–15}  The T12 vector, which harbors a truncation of the 5′ flanking region of HS4 (Figure 2Ai), showed significantly decreased globin expression both in vivo and in vitro in comparison to T9 (Figure 2Aii and 2C). The average vector copy number of vectors was stable throughout the course of the study (0.41 to 1.24) (Figure 2B). To further characterize the role of 5′ HS4, the 5′ flanking region of HS4 was replaced with different elements shown in Figure 2Ai. Only replacement of 5′ HS4 with the human IFN-β S/MAR²⁰  sequence in forward orientation restored average expression to the level obtained with the T9PD vector in MEL cells (Figure 2Aii), showing that HS flanking regions play an important role in LCR function and do not function solely as spacers between HS cores.

Mice treated with vectors containing the 5′HS1 (T10, n = 19 and S10, n = 17) exhibited the highest Hb levels, 129 (± 9) g/L and 127 (± 11g/L, respectively (Figure 2D). When normalized to vector copy, those vectors provide 95g/L and 88g/L of Hb per vector copy (Figure S1B). The total hemoglobin levels in mice treated with T9 (n = 19) and S9 (n = 18) were very similar (119 ± 8 g/L and 117 ± 6 g/L, respectively), and significantly lower than for T10 (n = 19) and S10 (n = 17; P = .002, Figure 2D). When normalized to vector copy, the T9 and S9 vectors provided approximately 64g/L and 42g/L of cHb per vector copy, respectively (Figure S1B). Addition of HS1 thus nearly doubled the hemoglobin output of S9. On the other hand, the T12 vector performed the least well, as mice treated with this vector exhibited significantly lower Hb levels (100 ± 5 g/L; P ≤ .001, compared with T9, Figure 2D), which translated into 9g cHb/L per vector (Figure S1B). Protein expression was directly proportional to β-globin mRNA transcript levels for all vectors as shown in Figure S1A (R² = 0.98). Figures 2E and S1B summarize the net hemoglobin increase in relation to vector copy number. These results indicate that 1 to 2 vector copies per cell achieve major therapeutic responses with vectors T9, S9, S10 and T10. In contrast, 6 copies per cell of the T12 vector would be required to treat anemia to the same extent as with 1 copy of T9.

Our data clearly demonstrate that HS1 significantly improves LCR function in vivo (P < .001; Figure 1C and Figure S1B). This fits with results of Pasceri et al,²¹  who found that HS1 enhanced LCR-dependent globin expression in nonviral constructs tested in transgenic mice. These findings were not predicted by MEL cell studies (Figure 1A and other studies^22,23 ), reinforcing the utmost importance of using in vivo disease models to assess complex regulatory elements and the design of therapeutic vectors.

Our studies also support an important role for the intact HS4 element. Truncation of 5′ HS4 indeed significantly decreased β-globin transgene expression at the mRNA (Figure 2C) and protein levels (Figure 2D,E and S1B). Intriguingly, addition of the human IFN-βS/MAR sequence in forward orientation (T12SAR2PD) restored LCR function to the level found in the parental T9 vector. This observation suggests that a functional S/MAR element may be present in the 5′flanking region of HS4, as previously suggested by Cunningham et al²⁴  Alternatively, the IFN-β S/MAR may rescue expression by anchoring proviral sequences to nuclear matrix and positioning them in transcriptional “hotspots.”²⁵ 

We therefore conclude that HS1 and HS4 are important contributors to achieving therapeutic expression from globin vectors, which directly affects the vector copy number that is required to treat the anemia. Indeed, more than 5 copies per cell are required when using the T12 vector, whereas 1 or 2 copies per cell are sufficient with T9, S9, S10 and T10. This dosage requirement is likely to bear on the safety of HSC transduction, which is thought to be increased by diminishing the vector copy number in single cells.¹⁷  The use of tissue-restricted vectors, which should greatly diminish the probability of trans-activating neighboring oncogenes in HSC, progenitor cells, and lymphocytes,¹⁸  is expected to further increase the safety of globin gene transfer.

Based on previous findings⁸  and those reported here, we are preparing a phase I clinical trial to investigate the safety and tolerability of globin gene transfer in subjects with β-thalassemia major who lack a matched donor.⁴  In this study, which was recently reviewed by the Recombinant DNA Advisory Committee (RAC; June 20, 2007; http://videocast.nih.gov/ram/rac062007.ram), autologous CD34⁺cells obtained after G-CSF mobilization will be transduced with the TNS9.3 vector and infused following nonmyeloablative conditioning. Should higher levels of globin expression be eventually needed, the S10 and T10 vectors described here will prove to be very useful.

We thank Dr Stefano Rivella for assistance with cloning the 5′HS1 element, Dr J. Bode for providing the IFNβ S/MAR, and Drs Beata Gajewska and Laurent Beloeil for help with animal work.

This work was supported in part by RO1 HL-57 612 and the Leonardo Giambrone Foundation for the Cure of Thalassemia.

Contribution: L.L. designed and executed experiments, and cowrote the manuscript; M.S. designed experiments and cowrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Michel Sadelain, Box 182, MSKCC, 1275 York Ave, New York, NY 10021; e-mail:m-sadelain@ski.mskcc.org.